mm: migrate: do not touch page->mem_cgroup of live pages
authorJohannes Weiner <hannes@cmpxchg.org>
Tue, 15 Mar 2016 21:57:19 +0000 (14:57 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 15 Mar 2016 23:55:16 +0000 (16:55 -0700)
commit6a93ca8fde3cfce0f00f02281139a377c83e8d8c
tree63846308d4ab27360afdb31330cb8dbf4251cbfe
parent23047a96d7cfcfca1a6d026ecaec526ea4803e9e
mm: migrate: do not touch page->mem_cgroup of live pages

Changing a page's memcg association complicates dealing with the page,
so we want to limit this as much as possible.  Page migration e.g.  does
not have to do that.  Just like page cache replacement, it can forcibly
charge a replacement page, and then uncharge the old page when it gets
freed.  Temporarily overcharging the cgroup by a single page is not an
issue in practice, and charging is so cheap nowadays that this is much
preferrable to the headache of messing with live pages.

The only place that still changes the page->mem_cgroup binding of live
pages is when pages move along with a task to another cgroup.  But that
path isolates the page from the LRU, takes the page lock, and the move
lock (lock_page_memcg()).  That means page->mem_cgroup is always stable
in callers that have the page isolated from the LRU or locked.  Lighter
unlocked paths, like writeback accounting, can use lock_page_memcg().

[akpm@linux-foundation.org: fix build]
[vdavydov@virtuozzo.com: fix lockdep splat]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/memcontrol.h
include/linux/mm.h
mm/filemap.c
mm/memcontrol.c
mm/migrate.c
mm/shmem.c